Cellular/Molecular

Scratch2 Prevents Cell Cycle Re-Entry by Repressing miR-25in Postmitotic Primary Neurons

Eva Rodríguez-Aznar, Alejandro Barrallo-Gimeno, and M. Angela NietoInstituto de Neurociencias, CSIC-UMH, San Juan de Alicante 03550, Spain

During the development of the nervous system the regulation of cell cycle, differentiation, and survival is tightly interlinked. Newly generatedneurons must keep cell cycle components under strict control, as cell cycle re-entry leads to neuronal degeneration and death. However, despitetheir relevance, the mechanisms controlling this process remain largely unexplored. Here we show that Scratch2 is involved in the control of thecell cycle in neurons in the developing spinal cord of the zebrafish embryo. scratch2 knockdown induces postmitotic neurons to re-enter mitosis.Scratch2 prevents cell cycle re-entry by maintaining high levels of the cycle inhibitor p57 through the downregulation of miR-25. Thus, Scratch2appears to safeguard the homeostasis of postmitotic primary neurons by preventing cell cycle re-entry.

IntroductionThe mechanisms controlling the cell cycle, cell differentiation,and survival are tightly regulated during nervous system devel-opment to ensure that the correct number of neurons and glialcells are generated when required (Guillemot, 2007). Neuronaldifferentiation is tightly linked to cell cycle withdrawal (Bally-Cuif and Hammerschmidt, 2003), and it has been classicallythought that neurons are locked in the postmitotic state. How-ever, postmitotic neurons must actively retain the control of theircell cycle components to prevent the reversion to the proliferativestate (Herrup and Yang, 2007; Deves and Bourrat, 2012). Inap-propriate cell cycle re-entry is associated with cell death and in-deed observed in a wide variety of pathological states, such asretinoblastoma and Alzheimer�s disease and in the response toDNA damage (Kruman et al., 2004; Herrup and Yang, 2007;Varvel et al., 2009). Thus, not only is it crucial to understand themechanisms that drive cell cycle exit but also those that preventits re-initiation, events, which are less well understood.

Cyclin-dependent kinase inhibitors (CKIs) and the retino-blastoma proteins are involved in the maintenance of the post-mitotic state (Sage et al., 2003; Ajioka et al., 2007; Laine et al.,2007; Pajalunga et al., 2007; Zhang et al., 2008). Therefore, it is

important to understand how the expression of CKIs is con-trolled both during and following neuronal differentiation.

Here we show that scratch2, a member of the Scratch family oftranscription factors specifically expressed in the nervous systemfrom Drosophila and Caenorhabditis elegans to vertebrates (Roark etal., 1995; Nakakura et al., 2001; Marín and Nieto, 2006; Rodríguez-Aznar and Nieto, 2011) maintains high levels of the CKI familymember p57 (cdkn1c), thereby preventing neuronal cell cycle re-entry. As such, we show that scratch2 knockdown induces prolifera-tion in differentiated neurons by increasing the expression of miR-25, which in turn, decreases p57 levels abrogating its function inmaintaining differentiated neurons in the postmitotic state.

Materials and MethodsFish maintenance. Wild-type (WT) strains used in this work are AB andTup-Lof zebrafish. The transgenic line Tg(huC:GFP) was kindly pro-vided by Elisa Martí. The transgenic line Tg(xEF1a:H2B-RFP) was gen-erated using the xEF1a minimal promoter driving the expression of redfluorescent protein mRFP1 fused in frame to the sequence of zebrafishH2B histone. This construct was introduced into one-cell zebrafish em-bryos using the miniTol2 vector, a kind gift from J.L. Gomez-Skarmeta(Balciunas et al., 2006) and Tol2 transposase mRNA. Embryos weremaintained at 28°C under standard conditions and staged as describedpreviously (Kimmel et al., 1995).

Isolation of cDNAs and PCR. Total RNA was extracted from 24 h post-fertilization (hpf) zebrafish embryos with Trizol (Invitrogen) and used asa template to synthesize cDNA with SuperScript III reverse transcriptase(Invitrogen) and oligodT as the primer. The following primers were usedfor cdkn1c (NM 001002040) forward: 5�-ggatccgctcacggcattgactttta-3�and reverse 5�-tctagatggactgatgtatgtatagtgcaaa-3�. The human Mcm7intronic region containing the miR-106b�25 cluster was amplified fromhuman DMS53 cells genomic DNA, using the primers: forward, 5�-ATGGATCCCTGGAGGGCCATCTGTTTT-3�;reverse,5�-GGTCTAGAGTGCCTAAGGGGAAGGTAGG-3�. Amplified fragments were subclonedinto pGEM.TE and pCS2�.

Morpholino oligonucleotides and mRNA injections. All morpholino an-tisense oligonucleotides were obtained from Gene Tools and used asdescribed by (Nasevicius and Ekker, 2000). The following morpholinooligonucleotides (MOs) were used at 2 ng/embryo: scratch2 MO 5�-CGCACCAGAAAATCCTTCACATCGG-3� located at the 5� UTR (described as

Received Sept. 19, 2012; revised Jan. 2, 2013; accepted Jan. 30, 2013.Author contributions: E.R.-A. and M.A.N. designed research; E.R.-A. performed research; A.B.-G. contributed

unpublished reagents/analytic tools; E.R.-A. and M.A.N. analyzed data; M.A.N. wrote the paper.Work in the lab is supported by grants from the Spanish Ministry of Science and Innovation (BFU2008-01042,

CONSOLIDER-INGENIO 2010 CSD2007-00017, and CSD2007-00023) and the Generalitat Valenciana (Prometeo 2008/049) to M.A.N. We are grateful to Diana Abad for excellent technical support in the fish facility, Giovanna Exposito forhelp with the confocal microscope, Stuart B. Ingham for his help with Figure 8 and video editing, and to all labmembers for their helpful discussions. We also thank Oscar Marin, Isabel Farinas, and Joan Galceran for helpfulcomments on this manuscript.

The authors declare no competing financial interests.Correspondence should be addressed to: M. Angela Nieto, Instituto de Neurociencias, CSIC-UMH, Avda. Ramon y

Cajal s/n, San Juan de Alicante 03550, Spain. E-mail: anieto@umh.es.A. Barrallo-Gimeno’s present address: Departament de Ciencies Fisiologiques II, Facultat de Medicina, Universitat

de Barcelona, Campus de Bellvitge, L�Hospitalet de Llobregat, 08907 Barcelona, Spain.DOI:10.1523/JNEUROSCI.4459-12.2013

Copyright © 2013 the authors 0270-6474/13/335095-11$15.00/0

The Journal of Neuroscience, March 20, 2013 • 33(12):5095–5105 • 5095

MO2 in Rodríguez-Aznar and Nieto, 2011) and miR-25 MO 5�-TGTCAGACCGAGACAAGTGCAATGC-3�. p53 MO: 5�-GCGCCATTTGCAAGAATTG-3� was injected at 4.5 ng/embryo (Robu et al., 2007; Rodríguez-Aznar and Nieto, 2011). The standard control morpholino from GeneTools was used for the control injections. All oligonucleotides were in-jected into the yolk of one- or two-cell embryos using a pressure injector(PicoSpritzer). For rescue and overexpression experiments mRNA wassynthesized and capped using the mMessage mMachine Kit (Ambion).Rescue analyses were performed with 75 pg/embryo of scratch2 mRNAlacking the 5�UTR region to which scratch2 MO binds. For overexpressionanalyses 100 pg/embryo of control (red fluorescent protein, rfp), 20 pg/embryo of cdkn1c, or 200 pg/embryo of hsa-miR25 mRNA were used.

In situ hybridization and immunohistochemistry. Antisense RNAprobes were generated from plasmids containing sequences for thedifferent genes. The scratch2 probe has been described previously(Rodríguez-Aznar and Nieto, 2011). Plasmid containing the huC se-quence was provided by M.J. Blanco, and those for islet1 (isl1), islet2(isl2), pax2a, and nkx6.1 by A. Barrallo-Gimeno. For immunohisto-chemistry we used the primary antibodies against as follows: pH3 (rabbitanti-pH3 antibody, 1:1000; Millipore Biotechnology) HuC/D (mouseanti HuC/D, 1:1000; Invitrogen), and green fluorescent protein (GFP)(rabbit anti-GFP, 1:1000; Invitrogen). The secondary antibodies usedwere biotinylated goat anti-mouse (1:1000; Vector Laboratories), biotin-ylated goat anti-rabbit (1:1000; Sigma), Alexa Fluor 568 goat anti-mouse,and Alexa Fluor 488 goat anti-rabbit (1/500) from Invitrogen. In situhybridization and immunohistochemistry of whole-mount zebrafishembryos were performed as described previously (Acloque et al., 2008).The embryos were then fixed and embedded in gelatin to obtain vi-bratome sections (30 �m). To visualize the nuclei, 1 �l of 1 mg/mlHoechst solution (bisbenzimide H; Sigma) was added on vibratome sec-tions in 1 ml of PBS. Embryos were photographed using an OlympusDP70 digital camera on either a Leica M10 dissecting microscope or aLeica DMR microscope. For fluorescently labeled embryos, confocal im-ages were acquired using a Leica TCS SP2 AOBS confocal microscope.

Quantitative real-time reverse transcription PCR. Quantitative real-time reverse transcription PCR (qRT-PCR) was performed using theStep One Plus (Applied Biosystems) sequence detection system and theSYBR Green method. The data represent the mean of three independentexperiments with each point analyzed in triplicate. Transcript levels werecalculated using the comparative Ct method normalized to GAPDH. Thefinal results were expressed relative to the control condition using the2 �(��Ct) � SD formula. The primers used were as see in Table 1.

For mature dre-miR-25 detection, TaqMan PCR was performed usingthe TaqMan MicroRNA Cells-to-Ct Kit (Ambion) and hsa-miR-25 Taq-Man MicroRNA Assays (Applied Biosystems) primers. The results rep-resent the mean of at least three independent experiments �SD.

TUNEL assay. Apoptosis was analyzed using the TUNEL kit (Roche)incubated with the reagents according to manufacturer�s instructions asdescribed previously (Rodríguez-Aznar and Nieto, 2011).

Cell quantification. HuC/D- and pH3-positive cells were counted inconsecutive sections (30 �m thick) from embryos at 24 hpf correspond-ing to five consecutive somites (s5–s9). Fluorescently labeled cells werecounted in sections under the confocal microscope through z-scan. isl1-positive cells were counted in 24 hpf dorsal flat-mounted embryos.pax2a- and isl2-labeled cells were quantified in lateral flat-mounted em-bryos, in five hemisegments per embryo. Quantification of pax2a- andisl1-positive cells at the 16 somites stage spanned from s2 to s6.

Luciferase assays. To generate the luciferase reporter plasmids, a cdkn1c3� UTR fragment (407 bp from nucleotide 41 after the TGA stop codon tonucleotide 29 before the polyA sequence) containing the predicted

miR-25 binding site was subcloned into the pGL3-promoter vector (Pro-mega) downstream of the luciferase sequences. PCR amplification wasperformed on adult zebrafish genomic DNA as a template using theoligonucleotide primers 5�-GCAGAGCGGATGAAAGAAAG-3� and 5�-GATTCAAAGGTACAACGTGAGC-3�. Luciferase assays were per-formed in zebrafish embryos as described previously (Alcaraz-Perez etal., 2008). Briefly, the reporter plasmid of the firefly luciferase gene (25pg/e) was injected into one-cell embryos, together with 2.5 pg of a plas-mid containing the Renilla luciferase gene under the control of the Tkpromoter and the amount of the different MOs or mRNAs as describedin MOs and mRNA injections. Twenty embryos of each experimentalcondition were lysed in 100 �l of lysis buffer (Passive lysis buffer; Pro-mega) and assays were performed on 10 �l of the embryo lysate with aDual-Luciferase Reporter Assay System (Promega). The data are repre-sented as the mean � SD of three different experiments.

Live imaging and image processing. Transgenic embryos at 24 hpf weredechorioned and anesthetized before being mounted in 1% low meltingpoint agarose in 35 mm glass-bottom culture dishes (MatTek). Imageswere taken on the Leica TCS SP2 AOBS inverted confocal microscopeevery 3 �m, every 2.5 min with the 40� objective plus 2� optical zoom.Images were then processed with the Imaris software version 7.0 to pro-duce the videos. Twenty-four frames are shown over a region of 6 �m(corresponding to three confocal planes) at 3 frames per second; 3Dorthogonal representations of huC:GFP spinal cord vibratome sectionsshow a z-stack of 18 �m (seven confocal planes) using the Imaris soft-ware in Fib. 2B. Acquired images were further processed with AdobePhotoshop software to generate overlays.

Statistics. All values are presented as the mean � SD. All comparisonswere performed using the Student�s t test for samples with unequal vari-ance. Differences were considered statistically significant when p � 0.05;*p � 0.05, **p � 0.01, ***p � 0.001.

Database searches. NCBI and EMBL-EBI/Wellcome Trust Sanger In-stitute databases were searched using the BLAST family of programs(http://www.ncbi.nlm.nih.gov/BLAST) and http://www.ensembl.org,respectively, and the http://www.mirbase.org/, used for zebrafish cdkn1cmiRNA target prediction.

ResultsPostmitotic neurons in the zebrafish spinal cordexpress scratch2We previously demonstrated that zebrafish scratch genes are spe-cifically expressed in the developing nervous system and thatscratch2 is the only family member prominently expressed in thespinal cord (Rodríguez-Aznar and Nieto, 2011). We have nowcharacterized the scratch2-expressing cells and found that at 24hpf, it is confined to the mantle layer in the spinal cord (Fig.1A,B). We could not find cells that coexpressed scratch2 and themitotic cell marker phospho-histone 3 (pH3) (Fig. 1B). Indeeddouble-labeling experiments at 24 hpf indicated that scratch2-expressing cells are neurons that express the postmitotic markerhuC (Kim et al., 1996) and isl1, which is considered a marker ofearly neuronal differentiation and is expressed in Rohon Beard(RB), primary interneurons (INs), and motoneurons (MNs) atthe time of cell cycle exit (Korzh et al., 1993; Hutchinson andEisen, 2006) (Fig. 1C). scratch2 transcripts persisted at 36 hpf,although at much lower levels, as signal was only detected after amuch longer developing time (Fig. 1D). In summary, we find thatscratch2 is expressed in postmitotic spinal primary neurons.

The number of postmitotic neurons increases followingscratch2 knockdownTo understand the role of Scratch2 in neurogenesis, we disruptedits activity by injecting antisense MOs into one-cell zebrafish em-bryos. We previously used this strategy to demonstrate the im-portance of Scratch2 for the survival of spinal neurons and thus,the same experimental controls and conditions were used in this

Table 1. Primers

Gene Forward primer (5�-3�) Reverse primer (5�-3�)

Gapdh-AY818346 TGCGTTCGTCTCTGTAGATGT GCCTGTGGAGTGACACTGAcdkn1c-NM_001002040 CACATCCACGAGCATTGAAG CCGCTCTGCAGATAAACAGAmcm7-NM_212569 GACTACATCACTGCGGCGTA GGATGGACAGCAGTGTTCTGscratch2 As described previously (Rodríguez-Aznar and Nieto, 2011)

5096 • J. Neurosci., March 20, 2013 • 33(12):5095–5105 Rodríguez-Aznar et al. • Scratch2 Prevents Neuronal Cell Cycle Re-Entry

work (Rodríguez-Aznar and Nieto, 2011 and Materials andMethods). We next investigated whether scratch2 downregula-tion after morpholino injection affected the number of differen-tiating neurons by quantifying cells expressing HuC/D, usedextensively as a general marker for postmitotic neurons (Park etal., 2000). This approach allowed us to assess the overall impact ofscratch2 knockdown on the entire postmitotic neuronal popula-tion. We found that scratch2 morphants exhibited an increase inHuC/D-positive neurons compared with control embryos at 24hpf (Fig. 2A). This effect was specific to scratch2 loss of function,as coinjection of scratch2 MO with scratch2 mRNA reversed theincrease in HuC/D-positive neurons. scratch2 knockdown alsoleads to an upregulation of the p53 effector puma leading toneuronal apoptosis, which in turn induces proliferation of

neural progenitors in the ventricular zone(VZ; Rodríguez-Aznar and Nieto, 2011).Therefore, we wondered whether com-pensatory proliferation of the progenitorscould at least in part account for the neu-ronal hyperplasia phenotype observedin scratch2 morphants. We coinjectedscratch2 MO and p53 MO, as p53 inhibitionprevents the cell death induced by scratch2knockdown (Rodríguez-Aznar and Nieto,2011). As described, p53 MO injectionalone had no effect on embryonic devel-opment (Robu et al., 2007). Embryoscoinjected with both scratch2 and p53MOs (scratch2 MO � p53 MO) had moreHuC/D neurons like those injected withscratch2 MO alone (Fig. 2A). These resultssuggest that Scratch2 has an additionalfunction in the control of neuronal num-ber/differentiation independent from itsrole in promoting survival. To accuratelyquantify the increase in postmitotic neu-rons we analyzed serial confocal sectionsof scratch2 MO-injected embryos from aTg(huC:GFP) transgenic line (Fig. 2B)and after fluorescent immunohistochem-istry for HuC/D (Fig. 2C), and observedthat morphant embryos had 28% moreneurons than control embryos in bothcases (Fig. 2D; data not shown). The sameincrease was also observed with or with-out coinjection of p53 MO (Fig. 2D; datanot shown). Thus, Scratch2 has a specificrole in regulating the number of postmi-totic neurons independently from itsfunction in promoting neuronal survival.

The number of spinal primary neuronsis controlled by Scratch2In light of the data obtained, it was of in-terest to determine whether Scratch2 in-fluenced the number of all spinal neuronsor whether the increase in postmitoticneurons observed in the morphants wasassociated with specific populations. Theneuronal subpopulations in the spinalcord of 24 hpf zebrafish embryos includeprimary sensitive or RB neurons, INs andprimary MNs (Lewis and Eisen, 2003).

These neurons can be identified by the expression of differentmarkers: isl1 for RB, INs, and MNs (Korzh et al., 1993); pax2a forcommissural IN (Pfeffer et al., 1998); and isl2 for RB and primaryMN (Appel et al., 1995).

In scratch2 morphants at 24 hpf there was no change in thenumber of RB neurons identified by isl1 expression when com-pared with WT and controls (Fig. 3A; 6.15 � 0.24 in WT and6.2 � 0.53 in scratch2 MO cells per somite, n 7 embryos percondition). The number of INs analyzed by pax2a expression didnot change either (Fig. 3B; 4.03 � 0.78 in WT and 4.08 � 0.46 inscratch2 MO cells per hemisegment, n 6). In contrast, the num-ber of isl2-positive MNs was increased after scratch2 MO injection(Fig. 3C,D; 1.94 � 0.18 in WT and 4.67 � 0.6 cells per hemiseg-ment in scratch2 MO (p 1 � 10�6), n 6 embryos in control

Figure 1. Expression of scratch2 (scrt2) in the spinal cord of 24 hpf zebrafish embryos relative to that of other neural markers. A,Expression of scratch2 in a whole-mounted embryo at 24 hpf. B, scratch2 expression was not evident in mitotic cells revealed bypH3 staining in transverse sections at the mid spinal cord level. C, Coexpression of scratch2 with differentiation markers. scratch2is shown in red, other genes in green, and colocalization in yellow in the merged images. The markers selected were huC (postmi-totic) and isl1 (differentiated RB, primary INs and MNs). scratch2 colocalizes with both huC and isl1. D, scratch2 expressiondecreased in the 36 hpf spinal cord compared with that at 24 hpf (B). Embryo in D was overdeveloped to better assess expressionin the spinal cord. Scale bars, 10 �m.

Rodríguez-Aznar et al. • Scratch2 Prevents Neuronal Cell Cycle Re-Entry J. Neurosci., March 20, 2013 • 33(12):5095–5105 • 5097

and n 8 in scratcht2 morphants). As we observed a differentialresponse to scratch2 knockdown in the different populations, wewondered whether this might reflect the different timing of dif-ferentiation for the three subpopulations, and also whether it wascoupled with the temporal and spatial expression of scratch2.Indeed, scratch2 expression followed a progressive dorsoventralgradient in the developing spinal cord. At the 10 somites (14 hpf)scratch2 expression is restricted to the dorsal part of the spinalcord (Fig. 3E). At the 16 somites (17 hpf) stage expression isextended ventrally but still missing from the ventralmost regionin the spinal cord (Fig. 3F). By 24 hpf, transcripts are additionallydetected in MNs as reflected by their coexpression with nkx6.1(Cheesman et al., 2004) (Fig. 3G). Therefore, scratch2 knock-down may have differential effects on each subpopulation at dif-ferent times that follow the postmitotic state. In agreement withthis, we observed that at the 16 somite stage, the number of INswas increased in a similar manner to that observed in MNs inolder embryos (pax2a-positive cells 2.96 � 0.29 in WT and 3.8 �0.65 in scratch2 morphants; p 0.041, data not shown). How-ever, the number of MNs was not altered (1.84 � 0.36 in WT;1.8 � 0.24 in scratch2 MO, data not shown), as expected from theabsence of endogenous scratch2 expression at this stage (Fig. 3F).Together, our data demonstrate that scratch2 is involved in theregulation of neuronal number acting on postmitotic cells.

The scratch2 knockdown generates cycling cells in themarginal zoneOur previous results indicated that Scratch2, expressed in post-mitotic neurons, regulates the number of primary neurons, sug-

gesting that it influences the cell cycle. Hence, we assessed thenumber and position of the mitotic cells in the spinal cord ofembryos at 24 hpf, which, as described, were confined to the VZin control embryos (Fig. 4A; MOC, star). In contrast, a significantnumber of ectopically situated mitotic cells (double labeled forpH3 and HuC/D) were evident in scratch2 morphants (Fig. 4A,B;scrt2 MO, arrowhead; 88% 22/25). This phenotype was reversed bycoinjection of scratch2 mRNA, demonstrating that this effect wasspecific to scratch2 knockdown (Fig. 4A,B).

As the scratch2 MO does not induce its own mRNA degrada-tion, we can still detect scratch2 mRNA in the morphants, allowingus to identify the cells in which its function was downregulated.The cells containing scratch2 mRNA were those dividing ectopi-cally in the scratch2 morphants (Fig. 4C), suggestive of a cell-autonomous phenotype. To further analyze this phenotype, weused the Tg(huC:GFP) transgenic line and labeled all nuclei withHoechst staining. Most of these ectopically located mitotic cells inmorphants were huC positive (Fig. 4D). Similar results were ob-tained after giving a short bromodeoxyuridine pulse to livingTg(huC:GFP) embryos (data not shown). This indicates thatpostmitotic cells had re-entered the cell cycle. To confirm it invivo, we generated a H2B-RFP zebrafish transgenic line (see Ma-terials and Methods) to label all nuclei in red and cross the fishwith the Tg(huC:GFP) line that labels all postmitotic neurons ingreen. Time-lapse experiments on 24 hpf scrt2 morphant em-bryos revealed that some huC:GFP-positive cells in the mantlelayer entered mitosis and gave rise to two huC:GFP-labeleddaughter cells (Fig. 4E). In control embryos, all mitotic figureswere observed in the VZ. These results indicate that Scratch2

Figure 2. scratch2 knockdown induces neuronal hyperplasia in 24 hpf zebrafish embryos. Images show transverse sections of 24 hpf embryos at the spinal cord level. A, Increase in the numberof postmitotic neurons in scratch2 morphants is evident through the expression of HuC/D, a phenotype that was rescued by coinjection with scratch2 mRNA, indicating that the neuronal hyperplasiais specific to scratch2 knockdown. The increase in postmitotic neurons in the morphants was independent of neuronal death induced by scratch2 knockdown since blocking cell death by p53 MO

coinjection failed to rescue the neuronal hyperplasic phenotype. As expected, p53 morphant embryos were indistinguishable from WT embryos. B, Orthogonal projections of seven optical confocalplanes reconstructed with IMARIS software. scratch2 morphants display neuronal hyperplasia. C, Confocal planes of 24 hpf spinal cord transverse section labeled with antiHuC/D antibody. scratch2morphants display an increase in the number of postmitotic neurons. D, Quantification of HuC/D-positive cells in embryos such as those shown in C was performed under a confocal microscopethrough z-scans in five consecutive somites per embryo in six embryos per condition. The number of HuC/D-positive cells in each case was as follows: WT 29.9 � 1.75; scratch2 MO 38.5 � 2.07( p 0.00012); scratch2 MO � scratch2 29.7 � 1.7; p53 MO 30.7 � 1.7; scratch2 MO � p53 MO 38.76 � 1.95 ( p 7.1 � 10 �5). Scale bars: 10 �m.

5098 • J. Neurosci., March 20, 2013 • 33(12):5095–5105 Rodríguez-Aznar et al. • Scratch2 Prevents Neuronal Cell Cycle Re-Entry

function is required in postmitotic neurons to keep them in thepostmitotic state.

The expression of the p57 cell cycle inhibitor is enhancedby Scratch2Our previous data indicated that Scratch2 could act by prevent-ing cell cycle re-entry, suggesting that it may control componentsof the cell cycle machinery in postmitotic neurons. Indeed, ecto-pic cycling neurons positive for postmitotic differentiation mark-ers have been observed in mouse mutants for cell cycle regulators,including the p57 cell cycle inhibitor and retinoblastoma (Chenet al., 2007; Gui et al., 2007). The phenotype observed at 24 hpf inscratch2 morphants, with ectopic cycling neurons and an increasein MN number, is very similar to the that described after down-regulation of the cyclin-dependent inhibitor p57 in zebrafish 20hpf embryos (cdkn1c gene; Park et al., 2005). Moreover, we hadpreviously reported that the closely related Snail1 factor attenu-ates proliferation of epithelial cells by increasing the expression ofthe cell cycle inhibitor p21, another CKI family member (Vega etal., 2004; de Frutos et al., 2007). Considering the phenotypic

similarities and the fact that p57 is the only CKI described in theembryonic zebrafish spinal cord (Park et al., 2005), we studiedthe relationship between Scratch2 and p57. We observed that thelevel of cdkn1c transcripts diminished in 24 hpf embryos injectedwith scratch2 MO (Fig. 5A). Furthermore, p57 overexpression byinjection of cdkn1c mRNA was able to rescue the phenotype ofscratch2 morphants (Fig. 5B) as indicated by the reduction ofectopic pH3-positive cells (Fig. 5C). Altogether, these data indi-cate that the phenotype observed in scratch2 morphants is asso-ciated with aberrant low levels of p57, suggesting that Scratch2may increase p57 expression in vivo.

Scratch2 increases cdkn1c transcripts by downregulatingmiR-25 expressionWith a few exceptions, Scratch and Snail factors act as tran-scriptional repressors (Nakakura et al., 2001), suggesting thatthe induction of cdkn1c expression by Scratch2 might occurindirectly, perhaps through the repression of a p57 repressor. AsmicroRNAs (miRs) play an important role as cell cycle regulatorsby modulating the levels of CKIs (Bueno et al., 2008; Kim et al.,

Figure 3. Scratch2 regulates the number of spinal primary neurons. Lateral flat-mounted 24 hpf embryos shown anterior to the left. A, At 24 hpf, the number of RB-sensitive neurons determinedby isl1 expression in the dorsal spinal cord is similar in WT and scratch2 morphants. B, The number of commissural INs ( pax2a-positive cells) is not altered in scratch2 morphants. C, scratch2knockdown increases the number of MNs determined by isl2-positive cells (asterisk). D, Quantification of MNs from the experiment shown in C. The number of isl2-positive cells was 1.94 � 0.18 inWT and 4.67 � 0.6 in scratch2 MO ( p 1 � 10 �6) n 6 embryos in control embryos and n 8 in scratcht2 morphants. E–G scratch2 expression follows a temporal dorsoventral gradient in thedeveloping spinal cord. E, At the 10 somite stage, it is expressed in the dorsal spinal cord overlapping with RB markers. F, scratch2 expression in 16 somite stage embryos is absent from the ventralspinal cord (asterisk). G, By 24 hpf, scratch2 (red), is additionally coexpressed with MN markers such as nkx6.1 (green, asterisk). Scale bars: 10 �m.

Rodríguez-Aznar et al. • Scratch2 Prevents Neuronal Cell Cycle Re-Entry J. Neurosci., March 20, 2013 • 33(12):5095–5105 • 5099

2009) we wondered whether Scratch2could indirectly increase cdkn1c expres-sion by repressing a miR. To address thispossibility, we searched for miR targetsites in the cdkn1c 3�UTR region. Of thosefound, we focused on miR-25 since it canreduce cdkn1c mRNA levels in humancancer cells (Kim et al., 2009). miR-25 ishighly conserved among vertebrates, andit belongs to the miR-106b�25 cluster lo-cated in an intron of the mcm7 gene(Tanzer and Stadler, 2004; Aagaard et al.,2008).

We first investigated whether miR-25might regulate cdkn1c expression by ana-lyzing the activity of the zebrafish cdkn1c3�UTR sequence containing the miR-25binding site in a luciferase reporter vector.After injecting this construct into em-bryos together with the miR-106b�25cluster, luciferase activity was repressed at24 hpf, indicating that cdkn1c is a target ofmiR-25 in vivo (Fig. 6A). To confirm

Figure 4. Scratch2 prevents neuronal cell cycle re-entry. A, Transverse confocal images of 24 hpf zebrafish embryos showing mitotic cells (pH3 immunostaining, green) andHuC/D-positive neurons (red). In control embryos pH3-positive cells were located in the VZ (star), while in the scratch2 morphants (scratch2 MO) additional pH3-positive cells evident inthe mantle layer were also positive for HuC/D (arrowhead). These ectopic mitotic cells were specifically generated by scratch2 downregulation as coinjection of scratch2 mRNA rescuedthe phenotype (star). B, The number of ectopic mitotic cells was quantified in the different conditions, confirming the specificity of the scratch2 phenotype. Ten embryos were quantifiedper condition. Quantifications: WT 0.06 � 0.097; scratch2 MO 0.72 � 0.23 ( p 2.89 � 10 �6); scratch2 MO � scratch2 mRNA 0.22 � 0.17 ( p 5.16 � 10 �5 vs scratch2 MO).C, The presence of ectopic mitotic cells in the mantle layer was confirmed by the existence of scratch2/pH3-positive cells in the morphants (arrowhead). D, Tg(huC:GFP) embryos withnuclei stained for Hoechst (blue) confirm the presence of mitotic figures in huC-positive cells in scratch2 morphants (arrowhead), indicating that these embryos have defects in cell cycleregulation. HP, High-power magnification. E, Time-lapse analysis of living Tg(huC:GFP/xEF1a:H2B-RFP) double transgenic embryos showing cell divisions in huC-positive neurons inscratch2 morphants, anterior up. Snapshots taken from the same confocal plane at different time points show huC-positive cells (green) with red nuclei. White arrowhead points to agreen cell, where the nucleus undergoes mitosis, later giving rise to two huC:GFP-positive daughter cells. Scale bars: 10 �m.

Figure 5. Scratch2 increases cdkn1c expression. A, scratch2 morphants have reduced levels of cdkn1c transcripts as assessed byqPCR in 24 hpf embryos. Expression of cdkn1c in MOC: 1.12 � 0.17; scrt2 MO: 0.67 � 0.10, p 0.032 related to WT 1. n 3experiments. B, The aberrant presence of pH3-positive cells outside the VZ in the morphants is rescued by cdkn1c overexpression. C, Cellsdouble labeled for pH3 (green) and HuC/D (red) reflect the presence of differentiated neurons undergoing mitosis (arrowhead) unlikecontrol and rescue conditions (star). This phenotype is rescued in embryos coinjected with scratch2 MO and cdkn1c mRNA. Quantification:MOC: 0.08 � 0.11; scrt2 MO: 0.6 � 0.24 ( p vs MOC 0.0059); scrt2 MO�cdkn1c mRNA: 0.17 � 0.17 ( p vs scrt2 MO 0.0127). MOC andscrt2 MO embryos, n 5 per condition and scrt2 MO�cdkn1c mRNA embryos, n 7. Scale bars: 10 �m.

5100 • J. Neurosci., March 20, 2013 • 33(12):5095–5105 Rodríguez-Aznar et al. • Scratch2 Prevents Neuronal Cell Cycle Re-Entry

that the targeting of cdkn1c mRNA by miR-25 could havefunctional consequences, we analyzed the effect of miR-25overexpression in the embryos. We found that injection of themiR-25 cluster reduced the levels of cdkn1c transcripts in em-bryos analyzed at 24 hpf (Fig. 6B). The developing spinal cordscontained (1) cells double positive for pH3 and HuC/D lo-cated outside the VZ and (2) GFP-positive cells with nucleishowing mitotic figures when the miR-25 cluster was injectedinto Tg(huC:GFP) transgenic embryos (Fig. 6C,D). To betterassess whether cells divide after they have acquired the expres-sion of postmitotic markers, we followed the process in vivo byinjecting miR-25 mRNA into Tg(huC:GFP/xEF1a:H2B-RFP)double transgenic embryos. miR-25 overexpression resultedin the appearance of huC:GFP-positive cells entering mitosisand giving rise to two daughter cells (Fig. 6E), a phenotypereminiscent of that observed in scratch2 morphants.

The similarity between the two pheno-types suggests that miR-25 could mediatethe function of Scratch2 in the control ofneuronal number in the zebrafish spinalcord. To address this question, we first an-alyzed miR-25 expression levels in em-bryos injected with scratch2 MO and foundthat these morphants had higher levelsof mature miR-25 than the control em-bryos (Fig. 7A). This result is compatiblewith the decrease observed in cdkn1c expres-sion (Fig. 6B) and with the downregulationof miR-25 expression by Scratch2. However,the levels of mcm7, the host gene for miR-25,were not altered (Fig. 7B), suggesting thatmiR-25 and mcm7 may use independentpromoters, as suggested previously (Sikandet al., 2009; Monteys et al., 2010; Brett et al.,2011).

Our data suggest that Scratch2 preventspostmitotic neurons from re-entering thecell cycle by inhibiting miR-25 expression,which in turn leads to p57 activation. If true,the cell cycle re-entry that we had observedin postmitotic neurons in scratch2 mor-phants should be reverted upon miR-25downregulation. To decrease miR-25 in theembryos, we used miR-25MO injection. It isworth noting that miR-25 inhibition in-duces cell death, very likely because miR-25also targets p53 mRNA and that of the pro-apoptotic molecule bim (Petrocca et al.,2008; Kumar et al., 2011). Therefore, inthese experiments we also coinjected p53MO

to prevent cell death. This coinjectionstrategy, referred to as miR-25 MOs (s forsurvival), allowed us to assess the effectof miR-25 inhibition in the embryoswithout the interference produced byneuronal loss (Fig. 7C). We next con-firmed that miR-25 MO was attenuatingmiR-25 function. We injected it into ze-brafish embryos overexpressing miR-25and observed that it could rescue the ac-tivity of the luciferase reporter vectorcontaining the cdkn1c 3�UTR sequencewith the miR-25 binding site (Fig. 7D).

This indicates that miR-25 MO is functional in vivo.Injection of miR-25 MO in embryos prevented the appearance

of the ectopic cycling cells induced by scratch2 downregulation(Fig. 7E,F). This effect was accompanied by the restoration ofnormal cdkn1c transcript levels (Fig. 7G). Finally, the number ofisl2-positive MNs was rescued in 24 hpf embryos (Fig. 7H, I),demonstrating that supernumerary neurons found in scratch2morphants are the result of HuC/D cells that have re-entered thecell cycle and divided giving rise to two daughter cells. Together,our results indicate that Scratch2 downregulates miR-25 expres-sion, which in turn leads to p57 activation and prevents postmi-totic neurons from re-entering the cell cycle (Fig. 8).

DiscussionPostmitotic neurons must keep their cell cycle permanently incheck to prevent their reversion into a proliferative state

Figure 6. miR-25 overexpression induces HuC/D positive cells to re-enter mitosis. A, miR-25 represses a zebrafish 3�UTR cdkn1cluciferase (luc) reporter in vivo, indicating that cdkn1c is a miR-25 target gene in zebrafish. The results shown are the mean of threeindependent experiments, and measured relative to the control condition (value 1). miR-25 overexpression represses theactivity of p57 3�UTR by 40% (0.59 � 0.13; p 0.03). B, miR-25 overexpression is accompanied by a decrease in the levels ofcdkn1c transcripts. hsa-miR-25 0.74 � 0.21 related to MOC 1, p 0.17; n 3 experiments. C, D, The decrease in cdkn1cexpression observed after miR-25 overexpression was accompanied by the appearance of mitotic cells outside the VZ in 24 hpfembryos. As in scratch2 morphants, these cells are positive for both pH3 (green) and HuC/D (red) (arrowhead) in contrast to controlembryos (star). Note the appearance of irregular nuclei stained with Hoechst in D (arrowhead).The number of pH3-positive cellsoutside the VZ per somite increased in miR-25 overexpressing embryos (0.61 � 0.3 vs 0.086 � 0.11 in control embryos, p 5.67 � 10 �5). Quantification was performed on five control and 10 miR-25 overexpressing embryos. E, Time-lapse analysis ofliving Tg(huC:GFP/xEF1a:H2B-RFP) double transgenic embryos. Some huC-positive cells (green) enter mitosis and give rise to twodaughter cells (arrowhead). Images taken at a single focal plane at different time points. Scale bars: 10 �m.

Rodríguez-Aznar et al. • Scratch2 Prevents Neuronal Cell Cycle Re-Entry J. Neurosci., March 20, 2013 • 33(12):5095–5105 • 5101

Figure 7. Scratch2 prevents cell cycle re-entry by regulating miR-25 expression. A, B, miR-25 is upregulated in scrt2 morphants but mcm7, miR-25 host gene, does not appear altered.A, Expression of dre-miR-25 in scrt2 MO 1.42 � 0.069 related to MOC (1), p 0.01, n 3 experiments. B, Expression of mcm7 in scrt2 MO 1.04 � 0.24 related to MOC 1, n 3 experiments. C, miR-25 MO injection induces apoptosis in 92% of the embryos examined (n 30, arrowheads). This cell death can be prevented by coinjecting p53MO (miR-25 MOs,93.5%; n 46). Embryos coinjected with miR-25 MOs and scrt2 MO appeared as WT (81%; n 41). D, miR-25 MO is functional. miR-25 MO injection prevents the repression of the p57 3�UTRreporter observed after miR-25 overexpression. miR-25 represses the reporter by 40% (0.59 � 0.04, p 0.004) while miR-25 � miR-25 MO 0.86 � 0.12 ( p vs miR-25 0.044). Theresults shown are the mean of three independent experiments. E–G, miR-25 MO injection rescues the cell cycle phenotype observed in scratch2 morphants (star), both in terms ofpH3-positive cells outside the VZ double labeled with HuC/D or in Tg(huC:GFP) transgenic embryos with mitotic nuclei labeled with Hoechst (E, F, arrowhead, n 6) and cdkn1ctranscripts (G, n 4). Quantification in F: MOC 0.04 � 0.089; scrt2 MOs 0.74 � 0.1 p 2.08 � 10 �5; miR25 MO � p53 MO(miR25 MOs) 0.057 � 0.1; scrt2 MO � miR25 MOs 0.18 � 0.2 ( p 0.00068 vs scratch2 MOs). Quantification in G: MOC: 1; scrt2 MOs: 0.77 � 0,121 ( p 0.0132 vs MOC; p 0.0178 vs scrt2 MO � miR-25 MOs); miR-25 MOs: 1.1575 � 0.093( p 0.043 vs MOC); scrt2 MO � miR-25 MOs: 0.968 � 0.049. H, I, The increase in the number of isl2-expressing cells in scratch2 morphants (asterisks) reverted to that in WT embryoswhen miR-25 was downregulated. Quantification was performed in six embryos per condition. MOC 1.83 � 0.14; scrt2 MOs 3.75 � 0.51 p 0.00012; miR25 MO � p53 MO 1.87 �0.2; scrt2 MO � miR25 MOs 2.11 � 0.3. p compared with scrt2 MOs 0.00010. Scale bars: 10 �m.

5102 • J. Neurosci., March 20, 2013 • 33(12):5095–5105 Rodríguez-Aznar et al. • Scratch2 Prevents Neuronal Cell Cycle Re-Entry

(Herrup and Yang, 2007). Indeed, cell cycle re-entry is associ-ated with neuronal vulnerability (Herrup and Yang, 2007).Here, we show that Scratch2 regulates the cell cycle of spinalpostmitotic neurons in the developing zebrafish embryo.Scratch2 maintains the levels of the CKI family member p57,thereby preventing them from re-initiating proliferation andensuring neuronal homeostasis.

The increase in the number of neurons that we observed inscratch2 morphants is reminiscent of that observed in mutants forcell cycle components in the mouse (Chen et al., 2007; Gui et al.,2007) and in particular to that described in the spinal cord ofzebrafish cdkn1c morphants, where p57 is the only known CKIexpressed (Park et al., 2005). CKIs and p57 in particular, areexpressed in progenitors to arrest proliferation. Newborn p57mutant mice die soon after birth and display defects related totissue growth, delayed differentiation, and increased apoptosis inseveral organs that could be interpreted as the result of progeni-tors undergoing additional rounds of division before differenti-ation (Yan et al., 1997; Zhang et al., 1997; Georgia et al., 2006).Here, we demonstrate that scratch2 downregulation decreases thelevels of cdkn1c transcripts after neurons have exited the cell cycle,as Scratch2 is expressed in postmitotic neurons. Therefore, thephenotype observed after scratch2 knockdown is caused by post-mitotic cells re-entering the cell cycle while they continue to ex-press neuronal differentiation markers. Thus, independently ofthe role of p57 in promoting cell cycle withdrawal in committedprogenitors, Scratch2 specifically acts in postmitotic neurons,maintaining p57 expression high after cell cycle exit, preventingreversion to the proliferative state.

Inappropriate cell cycle re-entry has been associated with celldeath and indeed observed in a wide variety of animal models andpathological states such as neurodegenerative disorders (Lee etal., 1992; Dyer and Cepko, 2000; Zhang et al., 2008; reviewed byHerrup and Yang, 2007). These studies have shown that the ab-normal DNA replication that occurs when neurons can re-enterthe S-phase is rapidly followed by cell death. However, cell cyclere-entry does not seem to lead to neuronal death in scratch2 mor-phants or in embryos overexpressing miR-25, as we observe neu-rons able to progress into late G2-M phase (pH3 positive).Furthermore, live imaging analysis shows that aberrantly divid-ing postmitotic neurons complete mitosis and give rise to twodaughter cells both after scratch2 knockdown and miR-25 over-

expression. Therefore, it appears that dif-ferent cell types can respond differentlydepending on factors such as specific celldeath pathways, neuronal metabolism,epigenetic processes, or even morphol-ogy, which can make some cells moreprone to be reprogrammed than others(reviewed by Buttitta and Edgar, 2007 andDavis and Dyer, 2010). In addition to, cy-clinD expression decreases in mature cellsas differentiation proceeds (Skapek et al.,2001; Laine et al., 2010), which can makemore mature cells lacking cell cycle com-ponents be more prone to undergo celldeath than cell cycle re-entry.

An important question raised by ourinitial results was whether Scratch2 regu-lated the cell cycle independently of itsrole in the apoptotic pathway. We previ-ously found that aberrant apoptosis inscratch2 morphants induces compensa-

tory proliferation of the progenitors in an attempt to replenishthe damaged tissue (Rodríguez-Aznar and Nieto, 2011). By pre-venting cell death by downregulation of the p53 pathway, wedemonstrate here that the two pathways can act independentlyalbeit simultaneously, confirming that coordinating the survivaland the proliferation block mediated by Scratch2 contributes tothe final number of neurons.

Our data also indicate that the influence of Scracth2 on theneuronal differentiation program is not associated with a partic-ular fate, as scratch2 downregulation affects the number of bothINs and MNs. By analyzing the phenotype of scratch2 morphantsat different developmental stages, we have revealed thatScratch2 acts along the dorsoventral gradient of differentiation,evident through its sequential onset of expression in INs andMNs in a cell-autonomous manner.

A key finding of the present study is that Scratch2 controls p57expression through the downregulation of miR-25, throughwhich Scratch2 prevents neuronal cell cycle re-entry. These resultsexpand the importance of miRs in embryonic development,ranging from the definition of embryonic territories (Spruce etal., 2010) to neuronal differentiation and homeostasis. Interest-ingly, miR-9 was recently shown to regulate cell cycle exit andapoptosis in the Xenopus nervous system by downregulating theexpression of the hairy1 transcription factor in neuronal progen-itors (Bonev et al., 2011). They show that miR-9 is required in theprogenitors for cell cycle exit, and we show here that miR-25subsequently acts in postmitotic neurons to prevent them fromre-initiating proliferation. Thus, in addition to the sequentialspecification of neural fates by transcription factor codes (Guil-lemot, 2007), it appears that a superimposed sequential use ofdifferent miRs, including miR-25 and miR-9, adds a new level ofregulation to the neural differentiation program.

Our observations in the developing zebrafish spinal cord arelikely to be relevant to other systems, not least because miR-25 ishighly conserved in vertebrates and p57 has been implicated inregulating the cell cycle of neural precursors in the mouse cere-bral cortex (Tury et al., 2011). The novel role described here forScratch2 and p57 in preventing cell cycle re-entry in postmitoticneurons may provide new insight into cell cycle control, not onlyduring embryonic development but also in adult neurons andfollowing its deregulation in pathological conditions. Interest-ingly, Scratch2 is expressed in distinct regions of the adult mouse

Figure 8. Scratch2 in primary neuron differentiation. Scratch2 is included in the neuronal differentiation program acting inpostmitotic neurons. It prevents cell cycle re-entry by indirectly maintaining high levels of the cell cycle inhibitor p57 through thedownregulation of miR-25 expression. When scratch2 is knocked down, increased miR-25 levels in postmitotic neurons represscdkn1c transcript levels allowing cell cycle re-entry and leading to an increase in the number of primary neurons.

Rodríguez-Aznar et al. • Scratch2 Prevents Neuronal Cell Cycle Re-Entry J. Neurosci., March 20, 2013 • 33(12):5095–5105 • 5103

brain (Marín and Nieto, 2006) and our preliminary experimentsin zebrafish indicate that it is present in wider areas, compatiblewith the fact that extensive neurogenesis has been described inthe adult zebrafish brain (Zupanc et al., 2005). For example, cellcycle components are deregulated in Alzheimer�s or Parkinson�sdiseases (Hoglinger et al., 2007; Varvel et al., 2009) and themiR106b�25 cluster is upregulated in human gastric cancer(Petrocca et al., 2008). Interestingly, miR-106b and miR-93, theother members of the miR106b�25 cluster, target p21 (Bueno etal., 2008; Kim et al., 2009) and we previously reported that Snail1,an important element in tumor progression closely related toScratch (Nieto, 2011), attenuates proliferation by increasing p21expression in different cell contexts (Vega et al., 2004; de Frutos etal., 2007).

Together, the present findings demonstrate that Scratch2, inaddition to protecting neurons from cell death by directly main-taining low levels of Puma (Rodríguez-Aznar and Nieto, 2011),prevents neuronal re-entry into the cell cycle by indirectly main-taining the levels of the cell cycle inhibitor p57 high. Thus,Scratch2 appears to safeguard the homeostasis of primary neu-rons acting through two independent albeit simultaneousmechanisms.

ReferencesAagaard LA, Zhang J, von Eije KJ, Li H, Saetrom P, Amarzguioui M, Rossi JJ

(2008) Engineering and optimization of the miR-106b cluster for ectopicexpression of multiplexed anti-HIV RNAs. Gene Ther 15:1536 –1549.CrossRef Medline

Acloque H, Wilkinson DG, Nieto MA (2008) In situ hybridization analysis ofchick embryos in whole-mount and tissue sections. Methods Cell Biol87:169 –185. CrossRef Medline

Ajioka I, Martins RA, Bayazitov IT, Donovan S, Johnson DA, Frase S, CiceroSA, Boyd K, Zakharenko SS, Dyer MA (2007) Differentiated horizontalinterneurons clonally expand to form metastatic retinoblastoma in mice.Cell 131:378 –390. CrossRef Medline

Alcaraz-Perez F, Mulero V, Cayuela ML (2008) Application of the dual-luciferase reporter assay to the analysis of promoter activity in Zebrafishembryos. BMC Biotechnol 8:81. CrossRef Medline

Appel B, Korzh V, Glasgow E, Thor S, Edlund T, Dawid IB, Eisen JS (1995)Motoneuron fate specification revealed by patterned LIM homeoboxgene expression in embryonic zebrafish. Development 121:4117– 4125.Medline

Balciunas D, Wangensteen KJ, Wilber A, Bell J, Geurts A, Sivasubbu S, WangX, Hackett PB, Largaespada DA, McIvor RS, Ekker SC (2006) Harness-ing a high cargo-capacity transposon for genetic applications in verte-brates. PLoS Genet 2:e169. CrossRef Medline

Bally-Cuif L, Hammerschmidt M (2003) Induction and patterning of neu-ronal development, and its connection to cell cycle control. Curr OpinNeurobiol 13:16 –25. CrossRef Medline

Bonev B, Pisco A, Papalopulu N (2011) MicroRNA-9 reveals regional diver-sity of neural progenitors along the anterior–posterior axis. Dev Cell 20:19 –32. CrossRef Medline

Brett JO, Renault VM, Rafalski VA, Webb AE, Brunet A (2011) The mi-croRNA cluster miR-106b�25 regulates adult neural stem/progenitorcell proliferation and neuronal differentiation. Aging 3:108 –124. Medline

Bueno MJ, Perez de Castro I, Malumbres M (2008) Control of cell prolifer-ation pathways by microRNAs. Cell Cycle 7:3143–3148. CrossRefMedline

Buttitta LA, Edgar BA (2007) Mechanisms controlling cell cycle exit uponterminal differentiation. Curr Opin Cell Biol 19:697–704. CrossRefMedline

Cheesman SE, Layden MJ, Von Ohlen T, Doe CQ, Eisen JS (2004) Zebrafishand fly Nkx6 proteins have similar CNS expression patterns and regulatemotoneuron formation. Development 131:5221–5232. CrossRef Medline

Chen D, Opavsky R, Pacal M, Tanimoto N, Wenzel P, Seeliger MW, Leone G,Bremner R (2007) Rb-mediated neuronal differentiation through cell-cycle-independent regulation of E2f3a. PLoS Biol 5:e179. CrossRefMedline

Davis DM, Dyer MA (2010) Retinal progenitor cells, differentiation, and

barriers to cell cycle reentry. Curr Top Dev Biol 93:175–188. CrossRefMedline

de Frutos CA, Vega S, Manzanares M, Flores JM, Huertas H, Martínez-FríasML, Nieto MA (2007) Snail1 is a transcriptional effector of FGFR3 sig-naling during chondrogenesis and achondroplasias. Dev Cell 13:872– 883.CrossRef Medline

Deves M, Bourrat F (2012) Transcriptional mechanisms of developmentalcell cycle arrest: problems and models. Semin Cell Dev Biol 23:290 –297.CrossRef Medline

Dyer MA, Cepko CL (2000) p57 kip2 regulates progenitor cell proliferationand amacrine interneuron development in the mouse retina. Develop-ment 127:3593–3605. Medline

Georgia S, Soliz R, Li M, Zhang P, Bhushan A (2006) p57 and Hes1 coordi-nate cell cycle exit with self-renewal of pancreatic progenitors. Dev Biol298:22–31. CrossRef Medline

Gui H, Li S, Matise MP (2007) A cell-autonomous requirement for Cip/Kipcyclin-kinase inhibitors in regulating neuronal cell cycle exit but not dif-ferentiation in the developing spinal cord. Dev Biol 301:14 –26. CrossRefMedline

Guillemot F (2007) Spatial and temporal specification of neural fates bytranscription factor codes. Development 134:3771–3780. CrossRefMedline

Herrup K, Yang Y (2007) Cell cycle regulation in the postmitotic neuron:oxymoron or new biology? Nat Rev Neurosci 8:368 –378. CrossRefMedline

Hoglinger GU, Breunig JJ, Depboylu C, Rouaux C, Michel PP, Alvarez-Fischer D, Boutillier AL, Degregori J, Oertel WH, Rakic P, Hirsch EC,Hunot S (2007) The pRb/E2F cell-cycle pathway mediates cell death inParkinson’s disease. Proc Natl Acad Sci U S A 104:3585–3590. CrossRefMedline

Hutchinson SA, Eisen JS (2006) Islet1 and Islet2 have equivalent abilities topromote motoneuron formation and to specify motoneuron subtypeidentity. Development 133:2137–2147. CrossRef Medline

Kim CH, Ueshima E, Muraoka O, Tanaka H, Yeo SY, Huh TL, Miki N (1996)Zebrafish elav/HuC homologue as a very early neuronal marker. NeurosciLett 216:109 –112. CrossRef Medline

Kim YK, Yu J, Han TS, Park SY, Namkoong B, Kim DH, Hur K, Yoo MW, LeeHJ, Yang HK, Kim VN (2009) Functional links between clustered mi-croRNAs: suppression of cell-cycle inhibitors by microRNA clusters ingastric cancer. Nucleic Acids Res 37:1672–1681. CrossRef Medline

Kimmel CB, Ballard WW, Kimmel SR, Ullmann B, Schilling TF (1995)Stages of embryonic development of the zebrafish. Dev Dyn 203:253–310.CrossRef Medline

Korzh V, Edlund T, Thor S (1993) Zebrafish primary neurons initiate ex-pression of the LIM homeodomain protein Isl-1 at the end of gastrulation.Development 118:417– 425. Medline

Kruman II, Wersto RP, Cardozo-Pelaez F, Smilenov L, Chan SL, Chrest FJ,Emokpae R Jr, Gorospe M, Mattson MP (2004) Cell cycle activationlinked to neuronal cell death initiated by DNA damage. Neuron 41:549 –561. CrossRef Medline

Kumar M, Lu Z, Takwi AA, Chen W, Callander NS, Ramos KS, Young KH, LiY (2011) Negative regulation of the tumor suppressor p53 gene by mi-croRNAs. Oncogene 30:843– 853. CrossRef Medline

Laine H, Doetzlhofer A, Mantela J, Ylikoski J, Laiho M, Roussel MF, Segil N,Pirvola U (2007) p19(Ink4d) and p21(Cip1) collaborate to maintain thepostmitotic state of auditory hair cells, their codeletion leading to DNAdamage and p53-mediated apoptosis. J Neurosci 27:1434 –1444. CrossRefMedline

Laine H, Sulg M, Kirjavainen A, Pirvola U (2010) Cell cycle regulation in theinner ear sensory epithelia: role of cyclin D1 and cyclin-dependent kinaseinhibitors. Dev Biol 337:134 –146. CrossRef Medline

Lee EY, Chang CY, Hu N, Wang YC, Lai CC, Herrup K, Lee WH, Bradley A(1992) Mice deficient for Rb are nonviable and show defects in neuro-genesis and haematopoiesis. Nature 359:288 –294. CrossRef Medline

Lewis KE, Eisen JS (2003) From cells to circuits: development of the ze-brafish spinal cord. Prog Neurobiol 69:419 – 449. CrossRef Medline

Marín F, Nieto MA (2006) The expression of Scratch genes in the develop-ing and adult brain. Dev Dyn 235:2586 –2591. CrossRef Medline

Monteys AM, Spengler RM, Wan J, Tecedor L, Lennox KA, Xing Y, DavidsonBL (2010) Structure and activity of putative intronic miRNA promoters.RNA 16:495–505. CrossRef Medline

Nakakura EK, Watkins DN, Schuebel KE, Sriuranpong V, Borges MW, Nel-

5104 • J. Neurosci., March 20, 2013 • 33(12):5095–5105 Rodríguez-Aznar et al. • Scratch2 Prevents Neuronal Cell Cycle Re-Entry

kin BD, Ball DW (2001) Mammalian Scratch: a neural-specific Snailfamily transcriptional repressor. Proc Natl Acad Sci U S A 98:4010 – 4015.CrossRef Medline

Nasevicius A, Ekker SC (2000) Effective targeted gene ‘knockdown’ in ze-brafish. Nat Genet 26:216 –220. CrossRef Medline

Nieto MA (2011) The ins and outs of the epithelial to mesenchymal transi-tion in health and disease. Annu Rev Cell Dev Biol 27:347–376. CrossRefMedline

Pajalunga D, Mazzola A, Salzano AM, Biferi MG, De Luca G, Crescenzi M(2007) Critical requirement for cell cycle inhibitors in sustaining non-proliferative states. J Cell Biol 176:807– 818. CrossRef Medline

Park HC, Hong SK, Kim HS, Kim SH, Yoon EJ, Kim CH, Miki N, Huh TL(2000) Structural comparison of zebrafish Elav/Hu and their differentialexpressions during neurogenesis. Neurosci Lett 279:81– 84. CrossRefMedline

Park HC, Boyce J, Shin J, Appel B (2005) Oligodendrocyte specification inzebrafish requires notch-regulated cyclin-dependent kinase inhibitorfunction. J Neurosci 25:6836 – 6844. CrossRef Medline

Petrocca F, Vecchione A, Croce CM (2008) Emerging role of miR-106b-25/miR-17–92 clusters in the control of transforming growth factor betasignaling. Cancer Res 68:8191– 8194. CrossRef Medline

Pfeffer PL, Gerster T, Lun K, Brand M, Busslinger M (1998) Characteriza-tion of three novel members of the zebrafish Pax2/5/8 family: dependencyof Pax5 and Pax8 expression on the Pax2.1 (noi) function. Development125:3063–3074. Medline

Roark M, Sturtevant MA, Emery J, Vaessin H, Grell E, Bier E (1995) scratch,a pan-neural gene encoding a zinc finger protein related to snail, pro-motes neuronal development. Genes Dev 9:2384 –2398. CrossRefMedline

Robu ME, Larson JD, Nasevicius A, Beiraghi S, Brenner C, Farber SA, EkkerSC (2007) p53 activation by knockdown technologies. PLoS Genet3:e78. CrossRef Medline

Rodríguez-Aznar E, Nieto MA (2011) Repression of Puma by scratch2 isrequired for neuronal survival during embryonic development. CellDeath Differ 18:1196 –1207. CrossRef Medline

Sage J, Miller AL, Perez-Mancera PA, Wysocki JM, Jacks T (2003) Acutemutation of retinoblastoma gene function is sufficient for cell cycle re-entry. Nature 424:223–228. CrossRef Medline

Sikand K, Slane SD, Shukla GC (2009) Intrinsic expression of host genes and

intronic miRNAs in prostate carcinoma cells. Cancer Cell Int 9:21.CrossRef Medline

Skapek SX, Lin SC, Jablonski MM, McKeller RN, Tan M, Hu N, Lee EY(2001) Persistent expression of cyclin D1 disrupts normal photoreceptordifferentiation and retina development. Oncogene 20:6742– 6751.CrossRef Medline

Spruce T, Pernaute B, Di-Gregorio A, Cobb BS, Merkenschlager M, Man-zanares M, Rodriguez TA (2010) An early developmental role for miR-NAs in the maintenance of extraembryonic stem cells in the mouseembryo. Dev Cell 19:207–219. CrossRef Medline

Tanzer A, Stadler PF (2004) Molecular evolution of a microRNA cluster.J Mol Biol 339:327–335. CrossRef Medline

Tury A, Mairet-Coello G, DiCicco-Bloom E (2011) The cyclin-dependentkinase inhibitor p57Kip2 regulates cell cycle exit, differentiation, andmigration of embryonic cerebral cortical precursors. Cereb Cortex21:1840 –1856. CrossRef Medline

Varvel NH, Bhaskar K, Kounnas MZ, Wagner SL, Yang Y, Lamb BT, HerrupK (2009) NSAIDs prevent, but do not reverse, neuronal cell cycle reentryin a mouse model of Alzheimer disease. J Clin Invest 119:3692–3702.CrossRef Medline

Vega S, Morales AV, Ocana OH, Valdes F, Fabregat I, Nieto MA (2004) Snailblocks the cell cycle and confers resistance to cell death. Genes Dev 18:1131–1143. CrossRef Medline

Yan Y, Frisen J, Lee MH, Massagu e J, Barbacid M (1997) Ablation of theCDK inhibitor p57Kip2 results in increased apoptosis and delayed differ-entiation during mouse development. Genes Dev 11:973–983. CrossRefMedline

Zhang J, Cicero SA, Wang L, Romito-Digiacomo RR, Yang Y, Herrup K(2008) Nuclear localization of Cdk5 is a key determinant in the postmi-totic state of neurons. Proc Natl Acad Sci U S A 105:8772– 8777. CrossRefMedline

Zhang P, Liegeois NJ, Wong C, Finegold M, Hou H, Thompson JC, SilvermanA, Harper JW, DePinho RA, Elledge SJ (1997) Altered cell differentia-tion and proliferation in mice lacking p57KIP2 indicates a role inBeckwith-Wiedemann syndrome. Nature 387:151–158. CrossRefMedline

Zupanc GK, Hinsch K, Gage FH (2005) Proliferation, migration, neuronaldifferentiation, and long-term survival of new cells in the adult zebrafishbrain. J Comp Neurol 488:290 –319. CrossRef Medline

Rodríguez-Aznar et al. • Scratch2 Prevents Neuronal Cell Cycle Re-Entry J. Neurosci., March 20, 2013 • 33(12):5095–5105 • 5105